Near-Temperature-Independent Electron Transport Well beyond Expected Quantum Tunneling Range via Bacteriorhodopsin Multilayers

A key conundrum of biomolecular electronics is efficient electron transport (ETp) through solid-state junctions up to 10 nm, often without temperature activation. Such behavior challenges known charge transport mechanisms, especially via nonconjugated molecules such as proteins. Single-step, coherent quantum-mechanical tunneling proposed for ETp across small protein, 2–3 nm wide junctions, but it is problematic for larger proteins. Here we exploit the ability of bacteriorhodopsin (bR), a well-studied, 4–5 nm long membrane protein, to assemble into well-defined single and multiple bilayers, from ∼9 to 60 nm thick, to investigate ETp limits as a function of junction width. To ensure sufficient signal/noise, we use large area (∼10–3 cm2) Au–protein–Si junctions. Photoemission spectra indicate a wide energy separation between electrode Fermi and the nearest protein-energy levels, as expected for a polymer of mostly saturated components. Junction currents decreased exponentially with increasing junction width, with uniquely low length-decay constants (0.05–0.5 nm–1). Remarkably, even for the widest junctions, currents are nearly temperature-independent, completely so below 160 K. While, among other things, the lack of temperature-dependence excludes, hopping as a plausible mechanism, coherent quantum-mechanical tunneling over 60 nm is physically implausible. The results may be understood if ETp is limited by injection into one of the contacts, followed by more efficient charge propagation across the protein. Still, the electrostatics of the protein films further limit the number of charge carriers injected into the protein film. How electron transport across dozens of nanometers of protein layers is more efficient than injection defines a riddle, requiring further study.


Insight in AFM Scratching
The AFM scratching results of the T-60 bR multilayer (Figure S1B) shows incomplete removal of bR molecules at the scratched area, despite of the high applied contact force (~200 nN).Due to the incomplete removal of bR molecules, the estimated depth profile (Figure S1C) shows reduced thickness compared to the thickness (60 nm) deduced from ellipsometry.
To derive the overall layer thickness, we used a technique of AFM-scratched image analysis based on color-mapped height profiles (Figures S4A-D) for different bR multiple bilayers, where the minimum height of exposed protein layer was set to be zero (by zero correction using Gwyddion 2.63).The height corresponding to the color of the unscratched region, shows the layer thickness associated with the color bar.In Figures S4A-D analysis, green represents the average thickness of each bilayer, which nicely varies from one layer to another.In addition, area-based histogram analysis was also done on the scratched images, which essentially showed two height peaks (Figure S4E).
One peak corresponds to the average lowest-height exposed area (scratched region) and the other one is the average height of the unscratched region.Therefore, the difference of the height peaks, directly gives the layer thickness.

Setup to Maintain High Humidity Environment for Impedance Measurement
The protein junction was kept in a humidity chamber for 24h, which results in the appearance of several tiny droplets over the substrate (> 95% RH).Then, we transferred the humidity treated junction in the Lakeshore chamber and measured the impedance immediately.Before transferring the sample to the Lakeshore chamber (in probe station), we put a small Petri dish with wet filter paper and kept it for few hours with closed chamber lid.In this setup we maintained the high humidity; here the tiny droplet acts as an indicator that persists during the measurement.Moreover, without nitrogen drying, such protein junction cannot be recovered to dry protein junction characteristic (semicircular Nyquist plot) even after 3-4 h.

HOMO-LUMO Gap of bR
The allowed energy levels of bR were derived from the UV-vis of bR.The UV-vis absorption shows the characteristic absorption peaks ~280 nm and ~578 nm due to the aromatic amino acid residues (of the polypeptide) and the retinal part of bR, respectively.
The absorption edge at the longest wavelength (~650 nm) corresponds to the HOMO-LUMO gap of bR with equivalent energy ~1.9 eV.The shorter wavelength absorption edge (~460 nm) of the retinal corresponds to the energy gap of 2.7 eV relative to the HOMO.The nearly-forbidden energy gap of ~1.3 eV lies between the longer wavelength absorption edge (~310 nm; 4.0 eV) of polypeptide energy levels (violet line in Figure S10 right) and the shorter absorption edge of the retinal.

Simmons tunneling under low bias (
Here, I is the junction current under the applied bias, V; r is the protein layer thickness, φ is the energy height of the barrier, me the mass of the electron, and ħ = h/2π, is the reduced Planck constant. The equation S1a says ln (I) should vary linearly with ln (V) with a slope (= 1) in the low bias regime (V → 0).The experimental I-V curves closely fit for the junctions of just silicon oxide and linker-coated silicon oxide with a slope (~1.1).On the contrary, the fitted slopes largely deviate from 1 for every bR bilayer junction.Interestingly, for all bR bilayers the slopes are nearly identical (~0.1) (Figure S15), irrespective of protein-layer thickness.PS I, 3 Az, 4,5 and BSA 4 data are taken from our previous work.

FigureFigure S2 :Figure S4 :Figure S5 :
Figure S1: (A) Tapping mode AFM topography of 60 nm (ellipsometry) thick bR multilayer (T-60) on top of the linker-coated silicon (rms roughness ~5 nm).(B) Topography of tapping mode AFM image over scratched area of T-60 bR layer (C) Depth profile (white dotted line shown in (B) at the scratched region with the thickness ~50 nm.

Figure S9 :
Figure S9:A plot of length decay constant (β value @ 0.1 V) against the thickness of mono (or, for bR, bi-)layers of different proteins, including all bR multilayer junctions.The data were obtained with the same p ++ -Si/SiOx/Linker/Protein/Au-LOFO junction configurations.Ferritin,3 PS I,3 Az,4,5 and BSA 4 data are taken from our previous work.

Figure S11 :Figure S12A :
Figure S11: A plot of voltage breakdown with the protein layer thickness for Single (blue), Double (green), and Triple (red) bilayer junction, shows a constant electric field ~0.1 GV/m, which is required for electrical breakdown of different bR junctions.

Figure S12B :
Figure S12B: ln J vs. V plot of single bilayer bR junction on APTMS coated Si/SiOx with two different top electrodes at RT.The blue line represents data for the mechanically deposited Au-LOFO top electrode, for single bilayer junctions (± 0.5 V bias range), and the black line for thermally evaporated Pb-Au top electrodes; averaged over5 junctions (± 0.1 V bias range).

Figure S13 :Figure S14A :
Figure S13: Current-voltage (semi-log) response (used in NDC plot) of different bR bilayers including T-60 under the different applied bias and temperature conditions.(A) ± 0.1 V at 80K, (B) ± 0.1 V at 300K, (C) ± 0.5 V at RT, and (D) ±1.5 V at RT for double and triple bilayer, ± 1.3 V for single bilayer to avoid destroying the junction (voltage breakdown).No low temperature data was taken at the high voltage sweep as the junctions were not stable.

Figure S14B :
Figure S14B: Nyquist plots of single bilayer bR-junction under vacuum before exposure to high humidity (blue dots) and for the same sample re-dried, i.e., after removing it from the high humidity ambient, placing it back into vacuum and re-measuring it in vacuum after 1 day (yellow dots).The dots are the real data points and the black solid lines are the fits with the equivalent circuit, shown as inset (top-left), the same circuit as shown in Figure 9A as right-hand side inset.The values of the parameters for the fits to the data are shown in TableS3.

Table S1 :
Activation energies (Ea) for thermal activation of currents across p ++ -Si/SiOx/APTMS/bR/Au junctions with different bR-bilayers at ± 50 mV and ± 100 mV, applied bias both near RT and at low temperatures (<160K).TOP table (1A) for Ea estimated at negative bias and BOTTOM table (1B) for positive bias.